| Abhay Kumar | a61c522 | 2025-11-10 07:32:50 +0000 | [diff] [blame] | 1 | package runewidth |
| 2 | |
| 3 | import ( |
| 4 | "os" |
| 5 | "strings" |
| 6 | |
| 7 | "github.com/rivo/uniseg" |
| 8 | ) |
| 9 | |
| 10 | //go:generate go run script/generate.go |
| 11 | |
| 12 | var ( |
| 13 | // EastAsianWidth will be set true if the current locale is CJK |
| 14 | EastAsianWidth bool |
| 15 | |
| 16 | // StrictEmojiNeutral should be set false if handle broken fonts |
| 17 | StrictEmojiNeutral bool = true |
| 18 | |
| 19 | // DefaultCondition is a condition in current locale |
| 20 | DefaultCondition = &Condition{ |
| 21 | EastAsianWidth: false, |
| 22 | StrictEmojiNeutral: true, |
| 23 | } |
| 24 | ) |
| 25 | |
| 26 | func init() { |
| 27 | handleEnv() |
| 28 | } |
| 29 | |
| 30 | func handleEnv() { |
| 31 | env := os.Getenv("RUNEWIDTH_EASTASIAN") |
| 32 | if env == "" { |
| 33 | EastAsianWidth = IsEastAsian() |
| 34 | } else { |
| 35 | EastAsianWidth = env == "1" |
| 36 | } |
| 37 | // update DefaultCondition |
| 38 | if DefaultCondition.EastAsianWidth != EastAsianWidth { |
| 39 | DefaultCondition.EastAsianWidth = EastAsianWidth |
| 40 | if len(DefaultCondition.combinedLut) > 0 { |
| 41 | DefaultCondition.combinedLut = DefaultCondition.combinedLut[:0] |
| 42 | CreateLUT() |
| 43 | } |
| 44 | } |
| 45 | } |
| 46 | |
| 47 | type interval struct { |
| 48 | first rune |
| 49 | last rune |
| 50 | } |
| 51 | |
| 52 | type table []interval |
| 53 | |
| 54 | func inTables(r rune, ts ...table) bool { |
| 55 | for _, t := range ts { |
| 56 | if inTable(r, t) { |
| 57 | return true |
| 58 | } |
| 59 | } |
| 60 | return false |
| 61 | } |
| 62 | |
| 63 | func inTable(r rune, t table) bool { |
| 64 | if r < t[0].first { |
| 65 | return false |
| 66 | } |
| 67 | |
| 68 | bot := 0 |
| 69 | top := len(t) - 1 |
| 70 | for top >= bot { |
| 71 | mid := (bot + top) >> 1 |
| 72 | |
| 73 | switch { |
| 74 | case t[mid].last < r: |
| 75 | bot = mid + 1 |
| 76 | case t[mid].first > r: |
| 77 | top = mid - 1 |
| 78 | default: |
| 79 | return true |
| 80 | } |
| 81 | } |
| 82 | |
| 83 | return false |
| 84 | } |
| 85 | |
| 86 | var private = table{ |
| 87 | {0x00E000, 0x00F8FF}, {0x0F0000, 0x0FFFFD}, {0x100000, 0x10FFFD}, |
| 88 | } |
| 89 | |
| 90 | var nonprint = table{ |
| 91 | {0x0000, 0x001F}, {0x007F, 0x009F}, {0x00AD, 0x00AD}, |
| 92 | {0x070F, 0x070F}, {0x180B, 0x180E}, {0x200B, 0x200F}, |
| 93 | {0x2028, 0x202E}, {0x206A, 0x206F}, {0xD800, 0xDFFF}, |
| 94 | {0xFEFF, 0xFEFF}, {0xFFF9, 0xFFFB}, {0xFFFE, 0xFFFF}, |
| 95 | } |
| 96 | |
| 97 | // Condition have flag EastAsianWidth whether the current locale is CJK or not. |
| 98 | type Condition struct { |
| 99 | combinedLut []byte |
| 100 | EastAsianWidth bool |
| 101 | StrictEmojiNeutral bool |
| 102 | } |
| 103 | |
| 104 | // NewCondition return new instance of Condition which is current locale. |
| 105 | func NewCondition() *Condition { |
| 106 | return &Condition{ |
| 107 | EastAsianWidth: EastAsianWidth, |
| 108 | StrictEmojiNeutral: StrictEmojiNeutral, |
| 109 | } |
| 110 | } |
| 111 | |
| 112 | // RuneWidth returns the number of cells in r. |
| 113 | // See http://www.unicode.org/reports/tr11/ |
| 114 | func (c *Condition) RuneWidth(r rune) int { |
| 115 | if r < 0 || r > 0x10FFFF { |
| 116 | return 0 |
| 117 | } |
| 118 | if len(c.combinedLut) > 0 { |
| 119 | return int(c.combinedLut[r>>1]>>(uint(r&1)*4)) & 3 |
| 120 | } |
| 121 | // optimized version, verified by TestRuneWidthChecksums() |
| 122 | if !c.EastAsianWidth { |
| 123 | switch { |
| 124 | case r < 0x20: |
| 125 | return 0 |
| 126 | case (r >= 0x7F && r <= 0x9F) || r == 0xAD: // nonprint |
| 127 | return 0 |
| 128 | case r < 0x300: |
| 129 | return 1 |
| 130 | case inTable(r, narrow): |
| 131 | return 1 |
| 132 | case inTables(r, nonprint, combining): |
| 133 | return 0 |
| 134 | case inTable(r, doublewidth): |
| 135 | return 2 |
| 136 | default: |
| 137 | return 1 |
| 138 | } |
| 139 | } else { |
| 140 | switch { |
| 141 | case inTables(r, nonprint, combining): |
| 142 | return 0 |
| 143 | case inTable(r, narrow): |
| 144 | return 1 |
| 145 | case inTables(r, ambiguous, doublewidth): |
| 146 | return 2 |
| 147 | case !c.StrictEmojiNeutral && inTables(r, ambiguous, emoji, narrow): |
| 148 | return 2 |
| 149 | default: |
| 150 | return 1 |
| 151 | } |
| 152 | } |
| 153 | } |
| 154 | |
| 155 | // CreateLUT will create an in-memory lookup table of 557056 bytes for faster operation. |
| 156 | // This should not be called concurrently with other operations on c. |
| 157 | // If options in c is changed, CreateLUT should be called again. |
| 158 | func (c *Condition) CreateLUT() { |
| 159 | const max = 0x110000 |
| 160 | lut := c.combinedLut |
| 161 | if len(c.combinedLut) != 0 { |
| 162 | // Remove so we don't use it. |
| 163 | c.combinedLut = nil |
| 164 | } else { |
| 165 | lut = make([]byte, max/2) |
| 166 | } |
| 167 | for i := range lut { |
| 168 | i32 := int32(i * 2) |
| 169 | x0 := c.RuneWidth(i32) |
| 170 | x1 := c.RuneWidth(i32 + 1) |
| 171 | lut[i] = uint8(x0) | uint8(x1)<<4 |
| 172 | } |
| 173 | c.combinedLut = lut |
| 174 | } |
| 175 | |
| 176 | // StringWidth return width as you can see |
| 177 | func (c *Condition) StringWidth(s string) (width int) { |
| 178 | g := uniseg.NewGraphemes(s) |
| 179 | for g.Next() { |
| 180 | var chWidth int |
| 181 | for _, r := range g.Runes() { |
| 182 | chWidth = c.RuneWidth(r) |
| 183 | if chWidth > 0 { |
| 184 | break // Our best guess at this point is to use the width of the first non-zero-width rune. |
| 185 | } |
| 186 | } |
| 187 | width += chWidth |
| 188 | } |
| 189 | return |
| 190 | } |
| 191 | |
| 192 | // Truncate return string truncated with w cells |
| 193 | func (c *Condition) Truncate(s string, w int, tail string) string { |
| 194 | if c.StringWidth(s) <= w { |
| 195 | return s |
| 196 | } |
| 197 | w -= c.StringWidth(tail) |
| 198 | var width int |
| 199 | pos := len(s) |
| 200 | g := uniseg.NewGraphemes(s) |
| 201 | for g.Next() { |
| 202 | var chWidth int |
| 203 | for _, r := range g.Runes() { |
| 204 | chWidth = c.RuneWidth(r) |
| 205 | if chWidth > 0 { |
| 206 | break // See StringWidth() for details. |
| 207 | } |
| 208 | } |
| 209 | if width+chWidth > w { |
| 210 | pos, _ = g.Positions() |
| 211 | break |
| 212 | } |
| 213 | width += chWidth |
| 214 | } |
| 215 | return s[:pos] + tail |
| 216 | } |
| 217 | |
| 218 | // TruncateLeft cuts w cells from the beginning of the `s`. |
| 219 | func (c *Condition) TruncateLeft(s string, w int, prefix string) string { |
| 220 | if c.StringWidth(s) <= w { |
| 221 | return prefix |
| 222 | } |
| 223 | |
| 224 | var width int |
| 225 | pos := len(s) |
| 226 | |
| 227 | g := uniseg.NewGraphemes(s) |
| 228 | for g.Next() { |
| 229 | var chWidth int |
| 230 | for _, r := range g.Runes() { |
| 231 | chWidth = c.RuneWidth(r) |
| 232 | if chWidth > 0 { |
| 233 | break // See StringWidth() for details. |
| 234 | } |
| 235 | } |
| 236 | |
| 237 | if width+chWidth > w { |
| 238 | if width < w { |
| 239 | _, pos = g.Positions() |
| 240 | prefix += strings.Repeat(" ", width+chWidth-w) |
| 241 | } else { |
| 242 | pos, _ = g.Positions() |
| 243 | } |
| 244 | |
| 245 | break |
| 246 | } |
| 247 | |
| 248 | width += chWidth |
| 249 | } |
| 250 | |
| 251 | return prefix + s[pos:] |
| 252 | } |
| 253 | |
| 254 | // Wrap return string wrapped with w cells |
| 255 | func (c *Condition) Wrap(s string, w int) string { |
| 256 | width := 0 |
| 257 | out := "" |
| 258 | for _, r := range s { |
| 259 | cw := c.RuneWidth(r) |
| 260 | if r == '\n' { |
| 261 | out += string(r) |
| 262 | width = 0 |
| 263 | continue |
| 264 | } else if width+cw > w { |
| 265 | out += "\n" |
| 266 | width = 0 |
| 267 | out += string(r) |
| 268 | width += cw |
| 269 | continue |
| 270 | } |
| 271 | out += string(r) |
| 272 | width += cw |
| 273 | } |
| 274 | return out |
| 275 | } |
| 276 | |
| 277 | // FillLeft return string filled in left by spaces in w cells |
| 278 | func (c *Condition) FillLeft(s string, w int) string { |
| 279 | width := c.StringWidth(s) |
| 280 | count := w - width |
| 281 | if count > 0 { |
| 282 | b := make([]byte, count) |
| 283 | for i := range b { |
| 284 | b[i] = ' ' |
| 285 | } |
| 286 | return string(b) + s |
| 287 | } |
| 288 | return s |
| 289 | } |
| 290 | |
| 291 | // FillRight return string filled in left by spaces in w cells |
| 292 | func (c *Condition) FillRight(s string, w int) string { |
| 293 | width := c.StringWidth(s) |
| 294 | count := w - width |
| 295 | if count > 0 { |
| 296 | b := make([]byte, count) |
| 297 | for i := range b { |
| 298 | b[i] = ' ' |
| 299 | } |
| 300 | return s + string(b) |
| 301 | } |
| 302 | return s |
| 303 | } |
| 304 | |
| 305 | // RuneWidth returns the number of cells in r. |
| 306 | // See http://www.unicode.org/reports/tr11/ |
| 307 | func RuneWidth(r rune) int { |
| 308 | return DefaultCondition.RuneWidth(r) |
| 309 | } |
| 310 | |
| 311 | // IsAmbiguousWidth returns whether is ambiguous width or not. |
| 312 | func IsAmbiguousWidth(r rune) bool { |
| 313 | return inTables(r, private, ambiguous) |
| 314 | } |
| 315 | |
| 316 | // IsNeutralWidth returns whether is neutral width or not. |
| 317 | func IsNeutralWidth(r rune) bool { |
| 318 | return inTable(r, neutral) |
| 319 | } |
| 320 | |
| 321 | // StringWidth return width as you can see |
| 322 | func StringWidth(s string) (width int) { |
| 323 | return DefaultCondition.StringWidth(s) |
| 324 | } |
| 325 | |
| 326 | // Truncate return string truncated with w cells |
| 327 | func Truncate(s string, w int, tail string) string { |
| 328 | return DefaultCondition.Truncate(s, w, tail) |
| 329 | } |
| 330 | |
| 331 | // TruncateLeft cuts w cells from the beginning of the `s`. |
| 332 | func TruncateLeft(s string, w int, prefix string) string { |
| 333 | return DefaultCondition.TruncateLeft(s, w, prefix) |
| 334 | } |
| 335 | |
| 336 | // Wrap return string wrapped with w cells |
| 337 | func Wrap(s string, w int) string { |
| 338 | return DefaultCondition.Wrap(s, w) |
| 339 | } |
| 340 | |
| 341 | // FillLeft return string filled in left by spaces in w cells |
| 342 | func FillLeft(s string, w int) string { |
| 343 | return DefaultCondition.FillLeft(s, w) |
| 344 | } |
| 345 | |
| 346 | // FillRight return string filled in left by spaces in w cells |
| 347 | func FillRight(s string, w int) string { |
| 348 | return DefaultCondition.FillRight(s, w) |
| 349 | } |
| 350 | |
| 351 | // CreateLUT will create an in-memory lookup table of 557055 bytes for faster operation. |
| 352 | // This should not be called concurrently with other operations. |
| 353 | func CreateLUT() { |
| 354 | if len(DefaultCondition.combinedLut) > 0 { |
| 355 | return |
| 356 | } |
| 357 | DefaultCondition.CreateLUT() |
| 358 | } |